ROCKY MOUNTAIN

JOURNAL OF MATHEMATICS
Volume 34, Number 2, Summer 2004

NONDEGENERATE IDEALS IN
FORMAL POWER SERIES RINGS

CARLES BIVIÀ-AUSINA

ABSTRACT. We give a characterization of the ideals of fi-

nite colength of C[[x1 , . . . , xn ]] which have monomial integral
closure using concepts from local algebra, thus leading to an
algebraic proof of a result of Saia. We also construct a canon-
ical minimal reduction of any ideal in the class of monomial
ideals of C[[x1 , . . . , xn ]] of finite colength. Therefore, we give
an effective method for the computation of the multiplicity of
any ideal in this class.

1. Introduction. We denote by A the ring of formal power series

C[[x1 , . . . , xn ]]. By a result of Teissier [15], it is known that, when an
ideal I ⊆ A of finite colength (that is, dim C A/I < ∞) is generated by
monomials, then its multiplicity e(I) is equal to n!v(I), where v(I) is
the n-dimensional volume of the complementary in Rn+ of the Newton
polyhedron of I. The equality e(I) = n!v(I) is, in turn, equivalent to
say that the integral closure of I is generated by those monomials xk
such that k belongs to the Newton polyhedron of I. In this work we
characterize the ideals of C[[x1 , . . . , xn ]] satisfying this expression for
the multiplicity. Then we give an algebraic proof of the main result of
Saia in [13] that might be reproduced in more general contexts.
Motivated by the work of Yoshinaga [17] on the characterization of
Newton nondegenerate functions, in the sense of Kouchnirenko [7],
Saia established in [13] the definition of Newton nondegeneracy for
ideals in the ring of convergent power series On and proved that,
if I ⊆ On has finite colength, then I is Newton nondegenerate (or
simply, nondegenerate, for short) if and only if the integral closure of
I is determined by all the monomials with exponents in the Newton
polyhedron of I. The proof that Saia gives of this result deals with the
notion of toroidal embedding and the growth condition for the integral
2000 AMS Mathematics Subject Classification. Primary 14B05, Secondary
13H15.
Key words and phrases. Newton polyhedrons, integral closure, multiplicities.
Work supported by DGICYT Grant BFM2000-1110
Received by the editors on May 22, 2000, and in revised form on September 16,
2001.

c
Copyright
2004 Rocky Mountain Mathematics Consortium

495
496 C. BIVIÀ-AUSINA

closure, which is a characterization of the integral closure of ideals in

On through an analytic condition, see Theorem 2.4. Our approach to
determine the integral closure of nondegenerate ideals in A is based
on results from multiplicity theory in local rings and it can also be
reproduced for ideals in On of finite relative colength, which is a notion
defined in this work. The techniques introduced by Kouchnirenko in
[7], especially in the proof of [7, Theorem 2.8], are essential in our
work. Further developments of these techniques can be found in [1].
We remark that, using [13] and Rees’s theorem, see Theorem 2.6, the
equivalence between the equality e(I) = n!v(I) and the nondegeneracy
of I arises quite easily, see [1] for details.
To make the exposition clearer, we consider ideals of finite colength
in Sections 2 and 3. In Section 4 we show how to apply the ideas
of the previous sections to ideals of finite relative colength. In the
last section we use the notion of nondegeneracy to obtain a canonical
minimal reduction of any monomial ideal of A of finite colength, thus
giving an easy way to compute the multiplicity of a monomial ideal.

2. Nondegenerate ideals. We start by giving some elementary

definitions. As before, let A denote the ring of formal power series
C[[x1 , . . . , xn ]]. We set R+ = [0, +∞[, Z+ = Z ∩ R+ and Q+ =
Q ∩ R+ . If k = (k1 , . . . , kn ) ∈ Zn+ , we denote the monomial xk11 · · · xknn
by xk .


Given an element g = k ak xk ∈ A, the support of g is defined as
supp (g) = {k ∈ Zn+ : ak = 0}. If I ⊆ A is an ideal, we define the
support of I as the union of the supports of the elements of I and we
denote this subset of Zn+ by supp (I).

Definition 2.1. We say that a subset Γ+ of Rn+ is a Newton

polyhedron when there exists some A ⊆ Qn+ such that Γ+ is equal
to the convex hull in Rn+ of the set {k + v : k ∈ A, v ∈ Rn+ }. In this
case we say that Γ+ is the Newton polyhedron determined by A and
it is also denoted by Γ+ (A). If g ∈ A, the Newton polyhedron of g
is Γ+ (g) = Γ+ (supp (g)). Let I be an ideal of A; then the Newton
polyhedron of I is Γ+ (I) = Γ+ (supp (I)).
We observe that if g1 , . . . , gs is a generating system of I ⊆ A, then
Γ+ (I) is equal to the convex hull of Γ+ (g1 ) ∪ · · · ∪ Γ+ (gs ). It is also
 NONDEGENERATE IDEALS 497

obvious that, if J ⊆ I are two ideals of A, then Γ+ (J) ⊆ Γ+ (I).

Let Γ+ ⊆ Rn+ be a Newton polyhedron and let v ∈ Rn+ ; then we
define
l(v, Γ+ ) = min{v, k : k ∈ Γ+ };
∆(v, Γ+ ) = {k ∈ Γ+ : v, k = l(v, Γ+ )}.

Then a face of Γ+ is any set of the form ∆(v, Γ+ ) for some v ∈ Rn+ .
The dimension of a face ∆ of Γ+ , denoted by dim ∆, is defined as
the dimension of the smallest affine subspace of Rn+ containing ∆. We
denote by Γ the union of the compact faces of Γ+ and by Γ− the union
of all segments joining the origin and a point of Γ. If I is an ideal of
A, the sets Γ(I) and Γ− (I) are defined analogously.
We observe that, when I is an ideal of A of finite colength, then its
Newton polyhedron intersects all coordinate axes, since I contains a
power of the maximal ideal of A. This implies that the n-dimensional
volume of Γ− (I) is not zero. We will denote this volume by v(I),
and we observe that v(I) = Voln (Rn+ \ Γ+ (I)), where Voln denotes
n-dimensional volume.
Let ∆ be a compact face of a Newton polyhedron Γ+ ; then C(∆)
shall denote the cone obtained by all half lines starting at the origin
and passing through a point of ∆. Then we denote by A∆ the subring
of A given by all h ∈ A such that supp (h) ⊆ C(∆). Moreover, the
ring A∆ can be expressed as A∆ = A/P∆ , where P∆ is the ideal of
A generated by all the monomials xk such that k ∈ / C(∆). Hence we
have that A∆ is a complete local ring of dimension dim ∆ + 1, see [2,
p. 210].


If g = k ak xk ∈ A, we denote by g∆ the polynomial obtained by
the sum of all terms ak xk such that k ∈ supp (g) ∩ ∆.

Definition 2.2. Let I ⊆ A be an ideal; we say that I is Newton

nondegenerate or simply, nondegenerate if there exists a system of
generators g1 , . . . , gs of I such that, for every compact face ∆ of Γ+ (I),
the ideal of A∆ generated by (g1 )∆ , . . . , (gs )∆ has finite colength in
A∆ . In this case we say that {g1 , . . . , gs } is a nondegenerate system of
generators of I.

It is easy to check that if I ⊆ A is nondegenerate, then every

498 C. BIVIÀ-AUSINA

system of generators of I is nondegenerate. As can be seen in [7,

Theorem 6.2], the fact that (g1 )∆ , . . . , (gs )∆ has finite colength in A∆
is equivalent to say that, for all face ∆
⊆ ∆, the set of common zeros
of (g1 )∆
, . . . , (gs )∆
is contained in {x ∈ Cn : x1 · · · xn = 0}. It is
obvious that Definition 2.2 can be stated analogously for ideals in the
ring On of germs of analytic functions f : (Cn , 0) → C.
We say that an ideal of A is monomial when it is generated by
monomials. Observe that every monomial ideal of A is nondegenerate.
Let us denote the multiplicity, in the Samuel sense, of an ideal I ⊆ A
of finite colength by e(I), see [9] or [14] for elementary facts about
multiplicity theory in local rings. Now we state a result from [15] on
the multiplicity of a monomial ideal, see also [15, p. 340].

Lemma 2.3. Let I ⊆ A be a monomial ideal of finite colength; then

e(I) = n!v(I).

As we already mentioned in the introduction, the relation e(I) =

n!v(I) characterizes the class of nondegenerate ideals I in A of finite
colength. To show this, we need to introduce some concepts from
commutative algebra.
Let I be an ideal in an arbitrary ring R and h ∈ R. We say that
h is integral over I when there is a monic polynomial of the form
p(x) = xm + a1 xm−1 + · · · + am−1 x + am , with ai ∈ I i , m ≥ 1, such
that p(h) = 0. The subset of R of those elements which are integral
¯
over I is an ideal called the integral closure of I and it is denoted by I.
¯ ¯
The ideal I contains I and, if I = I, we say that I is integrally closed.
For the sake of completeness, we state a result of Lejeune-Teissier
[8] characterizing the integral closure of an ideal in On by an analytic
condition known as the growth condition for the integral closure, see
also [16, p. 326].

Theorem 2.4. Let I be an ideal of On and h ∈ On , then h is integral

over I if and only if there exists a constant C > 0 and a neighborhood
U of the origin in Cn such that
|h(x)| ≤ C sup{|gi (x)| : i = 1, . . . , s}, for all x ∈ U,
where g1 , . . . , gs is any generating system of I.
 NONDEGENERATE IDEALS 499

Remark 2.5. Given an arbitrary ideal I ⊆ A, we denote by I0

the ideal generated by all xk such that k ∈ Γ+ (I). It is clear that
I ⊆ I0 . When I is a monomial ideal, then its integral closure is
equal to I0 , see [4, p. 141]. Then for any ideal I ⊆ A, we have
¯ ⊆ Γ+ (I¯0 ) = Γ+ (I0 ) = Γ+ (I) and this gives the equality
that Γ+ (I)
¯
Γ+ (I) = Γ+ (I).
Given a pair of ideals I and J in a ring R such that J ⊆ I, we
say that J is a reduction of I when there exists an integer r > 0
such that I r+1 = JI r . As we shall see in the following lemma, the
notions of reduction, integral closure and multiplicity in a formally
equidimensional local ring (R, m) are closely related. Let e(I) denote
the multiplicity of an m-primary ideal I of R.

Theorem 2.6 [11]. Let (R, m) be a formally equidimensional local

ring, and let J and I be a pair of m-primary ideals of R such that
J ⊆ I. Then the following conditions are equivalent:
(1) J is a reduction of I;
(2) e(I) = e(J);
(3) I¯ = J.
¯

If I ⊆ A = C[[x1 , . . . , xn ]] has finite colength and J is a reduction

of I, then Γ+ (I) = Γ+ (J), as a consequence of the above theorem and
Remark 2.5.
Next we give a result from [10, p. 153] on the existence of reductions of
an ideal I in a local ring, see also [9, p. 112]. As we shall see, reductions
of an ideal can be obtained through generic linear combinations of a
generating system of I.

Theorem 2.7. Let R be an n-dimensional local ring whose residual

field is equal to C. Let I ⊆ R be an ideal of finite colength. Suppose that
I is generated by g1 , . . . , gs . Then there exists an analytic set S(I) ⊆
Cs × Cn such that,
if (a11 , . . . , a1s , . . . , an1 , . . . , ans ) ∈ Cs × Cn \ S(I)
and we define hi = j aij gj , i = 1, . . . , n, then the ideal of R generated
by h1 , . . . , hn is a reduction of I.

Now we show a pair of propositions, the proof of which are given in

500 C. BIVIÀ-AUSINA

Section 3 using the notion of Newton filtration.

Proposition 2.8. Let I ⊆ A be a nondegenerate ideal of finite

colength generated by n elements. Then
A
dim C = n!v(I).
I

Proposition 2.9. Let I ⊆ A be a nondegenerate ideal of finite codi-

mension and let J be a reduction of I. Then J is also nondegenerate.

Theorem 2.10. Let I be an ideal of finite colength in A. Then I is

nondegenerate if and only if e(I) = n!v(I).

Proof. Suppose that I is nondegenerate. Let us fix a system of

generators g1 , . . . , gs of I. If ∆ is any compact face of Γ+ (I), we denote
by I∆ the ideal of A∆ generated by (g1 )∆ , . . . , (gs )∆ .
By Theorem 2.7, we can find an analytic subset S ⊆ Cs × Cn such
. . . , ans ) ∈ Cs × Cn

that whenever (a11 , . . . , a1s , . . . , an1 ,
\ S, then the
ideal J generated by the series h1 = j a1j gj , . . . , hn = j anj gj is a
reduction of I and the ideal J∆ of A∆ generated by (h1 )∆ , . . . , (hn )∆
is a reduction of I∆ for any compact face ∆ of Γ+ (I). We observe that,
keeping the notation introduced in Theorem 2.7, the set S can be taken
as the union of S(I) and the subsets S(I∆ ), where ∆ varies over the
compact faces of Γ+ (I).
Let ∆ ⊆ Γ(I) be a compact face, then J∆ has finite colength since J∆
is a reduction of I∆ . Therefore, J is a nondegenerate ideal generated
by n elements. But this implies that e(I) = e(J) = dim C A/J =
n!v(J) = n!v(I), by Theorem 2.6, Proposition 2.8 and the fact that
Γ+ (I) = Γ+ (J).
Suppose that e(I) = n!v(I); then by Lemma 2.3 and Theorem 2.6,
the ideal I is a reduction of I0 . Then, by Proposition 2.9 we have that
I is also nondegenerate.

Given an ideal I ⊆ A, we denote by KI the ideal generated by all

¯ The Newton polyhedron of KI is also
monomials xk belonging to I.
¯
denoted by C(I) in [13].
 NONDEGENERATE IDEALS 501

Corollary 2.11. Let I be an ideal of finite colength in A. Then the

equality e(I) = n!v(I) holds if and only if Γ+ (I) = Γ+ (KI ).

Proof. If e(I) = n!v(I), then I is a reduction of I0 by Lemma 2.3 and

Theorem 2.6. Then, I¯ = I¯0 and it follows that Γ+ (I) = Γ+ (KI ).
Suppose that Γ+ (I) = Γ+ (KI ), then v(I) = Voln (Rn+ \ Γ+ (KI )). But
we observe that, in general,
¯
n!Voln (Rn+ \ Γ+ (KI )) = n!v(KI ) = e(KI ) ≥ e(I)
= e(I) ≥ e(I0 ) = n!v(I).
Therefore, we obtain the equality e(I) = n!v(I).

From Theorem 2.10 and Corollary 2.11 we have that an ideal I ⊆ A

of finite colength is nondegenerate if and only if Γ+ (I) = Γ+ (KI ).
The same equivalence was obtained by Saia in [13] for ideals of finite
colength in the ring On of convergent power series using a different
approach based on the growth condition for the integral closure, see
Theorem 2.4. Therefore, we see that the membership test for integral
closure of monomial ideals shown in [3] can obviously also be applied
for the integral closure of nondegenerate ideals.

Example 2.12. Consider the ideal I of A = C[[x, y, z]] generated by

f1 = y b + y b1 z c1 + z c , f2 = xa + xa2 z c2 + z c , f3 = xa + xa3 y b3 + y b and
f4 = y b1 z c1 + xa2 z c2 + xa3 y b3 , where a, b, c, ai , bi , ci > 0. We suppose
that the points (0, b1 , c1 ), (a2 , 0, c2 ) and (a3 , b3 , 0) do not belong to the
half upper plane determined by the points (a, 0, 0), (0, b, 0) and (0, 0, c).
That is, the Newton polyhedron of I has the following form

q (0, 0, c)
A
 A
 A
 (0, b1 , c1 )
AqH
   HH
(a2 , 0, c2 ) qP
  Hq (0, b, 0)
 PPq
 (a3 , b3 , 0)

 

q

(a, 0, 0)
502 C. BIVIÀ-AUSINA

Then, by Theorem 2.10, the multiplicity of I is equal to e(I) =

3!v(I) = ab3 c2 + a3 bc1 + a2 b1 c + a3 b1 c2 + a2 b3 c1 .

3. The Newton filtration. As can be seen in Section 2 of [7], we

can associate a filtration in A to each Newton polyhedron Γ+ ⊆ Rn+
intersecting the coordinate axis. The family of cones C(∆), where
∆ varies in the set of compact faces of Γ+ , gives a decomposition of
Rn+ . Then we can consider the map h : Rn+ → R+ which is piecewise
linear on each cone C(∆) and whose restriction to Γ is identically 1.
We observe that it is possible to choose an integer M > 0 such that
M h(Nn ) ⊆ N. We take the minimum number M with this condition
and we consider the map φ : Rn+ → R+ given by φ = M h. Now for
any r ∈ N, we define the ideals

Ar = {g ∈ A : supp (g) ⊆ φ−1 (r + N)} ∪ {0}.

The collection {Ar }r≥0 is a filtration of A, that is, Ar As ⊆ Ar+s , for

all r, s ∈ N. Moreover, each ideal Ar is integrally closed and ∩r≥0 Ar =
{0}. This is known as the Newton filtration of A associated to Γ+ .
From this filtration, we can define the map d : A → N ∪ {+∞} given
by d(g) = max{r ∈ N : g ∈ Ar } = min{φ(k) : k ∈ supp (g)} when
g = 0 and d(0) = +∞. Then we have that Ar = {g ∈ A : d(g) ≥ r},
for all r ∈ N. If no confusion arises, we shall also refer to the map
d : A → N ∪ {+∞} as the Newton filtration associated to Γ+ (I).
From this filtration we can construct the graded ring A = gr A =
⊕r≥0 Ar , where Ar = Ar /Ar+1 , for all r ≥ 0. We define the Rees ring
of the filtration d as R( ) = ⊕∇∈Z A∇ ∇ ⊆ A[, −∞ ] where we set
Ar = A if r < 0. Then we have that A = R( )/−∞ R( ).

Definition 3.1. Given a series g ∈ A, we define the initial form of

g in (g), as the image of g in A = gr A, that is, in (g) = g + Ad(g)+1 ∈
Ad(g) . If I is any ideal of A, then we define the initial ideal of I,
denoted
by in (I), as the ideal of A generated by all in (f ), where f ∈ I.
If g = k ak xk ∈ A, we denote by p(g) the polynomial obtained by the
sum of those ak xk such that d(xk ) = d(g). This is called the principal
part of g with respect to Γ+ . Obviously, we have that in (g) = in (p(g)).
If ∆ is a compact face of Γ+ , then the Newton filtration associated to
Γ+ induces in a natural way a filtration in A∆ . We denote by R∆ ( )
 NONDEGENERATE IDEALS 503

the Rees ring associated to this filtration and by A∆ the graded ring
R∆ ( )/−∞ R∆ ( ). Now let π∆ : A → A∆ be the epimorphism defined
by π∆ (f + Ar ) = fC(∆) + (A∆ )r , where (A∆ )r = A∆ ∩ Ar , for all r ≥ 0,
and fC(∆) is the sum of all terms ak xk such that k ∈ supp (f ) ∩ C(∆)


whenever f = k ak xk .

Definition 3.2. If g ∈ A, we call the initial form of g over ∆ to

π∆ (in (g)) and we denote this element of A∆ by in∆ (g). If I ⊆ A is
any ideal, the initial ideal of I over ∆ is the ideal of A∆ generated by
all in∆ (g) where g ∈ I. This ideal is denoted by in∆ (I).

Remark 3.3. By the construction of the filtration, it follows that

d(xα xβ ) ≥ d(xα )+d(xβ ) for any monomials xα and xβ , and the equality
holds if and only if there exists a compact fact ∆ ⊆ Γ+ such that
α, β ∈ C(∆). This fact implies that either in (f )in (g) = in (f g) or
in (f ) in (g) = 0, for every f, g ∈ A, and that the last occurs if and only
if there exists no compact face ∆ ⊆ Γ+ such that supp (p(f )) ∩ C(∆) =
∅ and supp (p(g)) ∩ C(∆) = ∅.
On the other hand, it is not true that the ideal in∆ (I) is generated
by the initial forms over ∆ of a generating system of I.

Proposition 3.4. Let I be an ideal of A, and let g1 , . . . , gs be a

system of generators of I. Let ∆ be a compact face of Γ+ (I). Then the
ideal of A∆ generated by (g1 )∆ , . . . , (gs )∆ has finite colength if and only
if the ideal generated by in∆ (g1 ), . . . , in∆ (gs ) in A∆ has finite colength.

Proof. Let M be the value of the Newton filtration induced by

Γ+ (I) along the monomials with exponents in the boundary Γ(I).
Suppose that ∆ is a compact fact of Γ(I) and that the ideal I∆
of A∆ generated by (g1 )∆ , . . . , (gs )∆ has finite colength in A∆ .
Let r = dim A∆ , since the ring A∆ is local with residual field
equal to C, there is a sequence h1 , . . . , hr of linear combinations of
(g1 )∆ , . . . , (gs )∆ that is a system of parameters in A∆ . Moreover, by
[6], we know that the ring A∆ is Cohen-Macaulay, then h1 , . . . , hr
is a regular sequence in A∆ . In particular, h1 tM , . . . , hr tM is a
regular sequence in R∆ (d), which implies that in∆ (h1 ), . . . , in∆ (hr )
is a regular sequence in A∆ = R∆ (d)/t−1 R∆ (d). Since the ring
504 C. BIVIÀ-AUSINA

A∆ is also Cohen-Macaulay of dimension r, it follows that this se-

quence must generate an ideal of finite colength. We have de-
fined the elements hi as linear combinations of (g1 )∆ , . . . , (gs )∆ , thus
{in∆ (h1 ), . . . , in∆ (hr )}A∆ ⊆ {in∆ (g1 ), . . . , in∆ (gs )}A∆ , and this im-
plies that the ideal {in∆ (g1 ), . . . , in∆ (gs )}A∆ has also finite colength
in A∆ .
Suppose that the ideal in A∆ generated by in∆ (g1 ), . . . , in∆ (gs )
has finite colength. As above, consider a system of parameters
H1 , . . . , Hr of A∆ obtained as linear combinations

of the initial forms
in∆ (g1 ), . . . , in∆ (gs ). Suppose that Hi = j aij in∆ (gj ) where aij ∈ C
for all i = 1, . . . , r. Since A∆ is an r-dimensional Cohen-Macaulay ring,
we find that H1 , . . . , Hr is a regular sequence. In particular, the Koszul
complex associated to these elements is acyclic. Then we can apply
Theorem 4.1 of [7] to say that the graded rings gr (A∆ /h1 , . . . , hr )
and A∆ /H1
, . . . , Hr  are isomorphic as graded complex vector spaces,
where hi = j aij (gj )∆ , i = 1, . . . , r. Then the ideal in A∆ generated
by h1 , . . . , hr has finite colength in A∆ and I∆ has also finite colength.

Proof of Proposition 2.8. Let g1 , . . . , gn be a nondegenerate system

of generators of I. Let us consider the Koszul complex associated to
g1 , . . . , gn
n n n
(1) 0 −→ A(n) −→ A(n−1) −→ · · · −→ A( 1 ) −→ A.

Let A be the graded ring associated to the filtration d induced by

Γ+ (I). The above complex induces the graded complex of free A-
modules
n n n
(2) 0 −→ A(n) −→ A(n−1) −→ · · · −→ A( 1 ) −→ A.

We claim that, under the hypothesis of nondegeneracy, the complex (2)

is exact. To see this we observe that, from [7, Proposition 2.6], there
exists an exact graded complex with grade preserving morphisms of the
form
(3) 0 −→ A −→ Cn−1 −→ · · · −→ C1 −→ C0 −→ 0,

where Cl is the direct sum of those A∆ such that ∆ is a compact

l-dimensional face of Γ+ (I), for all l = 0, 1, . . . , n − 1.
 NONDEGENERATE IDEALS 505

Given a compact face ∆ of Γ+ (I), we can consider the Koszul complex

K∆ with coefficients in A∆ of in∆ (g1 ), . . . , in∆ (gn ). If we take the direct
sum of the complexes K∆ over all compact l-dimensional faces of Γ+ (I),
we obtain the complex

(n) ( n ) (n)
(4) 0 −→ Cl n −→ Cl n−1 −→ · · · −→ Cl 1 −→ Cl ,

for all l = 0, 1, . . . , n − 1. Then if we look at the diagram in [7, p. 15],

we find that the complex (2) is exact if the above complexes are exact
in dimensions ≥ n − l, for all l = 0, 1, . . . , n − 1, by a simple diagram
chase argument. But this is equivalent to say that the complex K∆
is exact in dimensions ≥ n − l for all l-dimensional faces ∆ and all
l = 0, 1, . . . , n − 1.
Let l ∈ {0, 1, . . . , n − 1}, and let ∆ be a compact l-dimensional face
of Γ+ (I). Then the ideal of A∆ generated by in∆ (g1 ), . . . , in∆ (gn ) has
finite colength, by Proposition 3.4. This implies that the depth in A∆ of
this ideal is equal to l +1, since A∆ is Cohen-Macaulay [6] of dimension
l + 1. Now by [9, Theorem 16.8], also known as the rigidity property of
the Koszul complex, the homology of K∆ is zero in dimensions ≥ n − l.
Therefore, the complex (2) is exact.
Let I
be the ideal of A generated by in (g1 ), . . . , in (gn ). By the
exactness of (2) we have that the Hilbert series of A/I
can be expressed
as H(t) = (1−tM )n HA (t), where HA (t) is the Hilbert series of A and M
is the value of d along the monomials with exponents in the boundary
Γ(I). Then if we now use [7, Lemma 2.9] and [7, Theorem 4.1], we find
that

A A
dim C = dim C
= lim (1−tM )n HA (t) = n!Voln (Γ− (I)) = n!v(I).
I I t→1

Proof of Proposition 2.9. Let us consider the Newton filtration

induced by Γ+ (I). We fix a nondegenerate system of generators
g1 , . . . , gs of I. Let J be a reduction of I; then there exists an r > 0 such
that I r+1 = JI r . In particular, we find that in (I)r+1 ⊆ in (I r+1 ) ⊆
in (J). But it is obvious that

({in (g1 ), . . . , in (gs )}A)r+1 ⊆ in (I)r+1 .

506 C. BIVIÀ-AUSINA

Then if ∆ is any compact face of Γ+ (I) and we apply π∆ to the above in-
clusion, we have that in∆ (J) contains the ideal ({in∆ (g1 ), . . . , in∆ (gs )}
A∆ )r+1 and therefore in∆ (J) has finite colength, by Proposition 3.4.
Since the graded ring A associated to the Newton filtration is
a Noetherian ring, there exist h1 , . . . , hl ∈ J such that its initial
forms in (h1 ), . . . , in (hl ) generate the ideal in (J). Then we have that
h1 , . . . , hl is a generating system of J by [4, Proposition 7.12]. To ap-
ply this proposition the ring A must be complete with respect to the
filtration induced by Γ+ (I). But, as can be seen in [12, p. 76], the com-
pleteness of A with respect to a filtration only depends on the radical
of the filtration (the common radical of each piece of the filtration),
which in this case is equal to the maximal ideal.
Then, given any compact face ∆ of Γ+ (I), if we apply the morphism
π∆ to the equality in (J) = {in (h1 ), . . . , in (hl )}A, we obtain that
in∆ (J) = {in∆ (h1 ), . . . , in∆ (hl )}A∆ . But, as we have seen before,
the ideal in∆ (J) has finite colength in A∆ . Then h1 , . . . , hl is a
nondegenerate system of generators of J by Proposition 3.4.

4. Ideals of finite relative colength. In this section we take ad-

vantage of the technique shown in the previous sections to characterize
the class of nondegenerate ideals of finite relative colength, which is a
wide class of ideals including the ideals of finite colength of A. We
recall that, if I is an ideal of A, then Γ(I) is the union of all compact
faces of the Newton polyhedron Γ+ (I).

Definition 4.1. Let I be an ideal of A, we set

Λ+ (I) = {λk : λ ≥ 0, k ∈ Γ(I)} ⊆ Rn+ ;

AI = {h ∈ A : supp (h) ⊆ Λ+ (I)};
c(I) = max{dim ∆ : ∆ is a compact face of Γ(I)}.

We observe that, in general, the set Λ+ (I) is not convex in Rn+ .

Nevertheless, it is not the case when n = 2. If Λ+ (I) is convex, then
Λ+ (I) ∩ Zn+ is a sub-semigroup of Zn+ and AI becomes a subring of A.
In fact, AI is a local ring of dimension c(I) + 1 in this case. We say
that I has finite relative colength when Λ+ (I) is convex and I admits
a generating system g1 , . . . , gs such that supp (g1 ) ∪ · · · ∪ supp (gs ) ⊆
 NONDEGENERATE IDEALS 507

Λ+ (I), and the ideal in AI generated by g1 , . . . , gs has finite colength

in AI .
We observe that Γ− (I) is contained in an affine subspace of dimension
c(I) + 1. Then we can consider the (c(I) + 1)-dimensional volume of
Γ− (I); we denote this volume also by v(I). When I is an ideal of A
of finite colength, then I has finite relative colength since the Newton
polyhedron of I intersects all the coordinate axes and consequently
AI = C[[x1 , . . . , xn ]] in this case.
If I ⊆ A is an ideal of finite relative colength, we can consider I as
an ideal of AI . Therefore, it makes sense to compute the multiplicity
of I in AI since AI is a local ring. We shall denote this multiplicity by
e(I, AI ). In the sequel, we will denote by I1 the ideal of AI generated
by all the monomials xk , with k ∈ Λ+ (I) ∩ Zn+ (compare with the
definition of I0 in Remark 2.5).
By analogy with Section 3, we can also consider a filtration on AI
associated to the Newton polyhedron Γ+ (I). We use this tool to prove
the following results.

Proposition 4.2. Let I ⊆ A be a nondegenerate ideal of finite

relative codimension. If J ⊆ A is a reduction of I, then J is also
nondegenerate.

Proof. Is analogous to the proof of Proposition 2.9.

Theorem 4.3. Let I ⊆ A be an ideal of finite relative colength.

Then I is nondegenerate if and only if e(I, AI ) = (c(I) + 1)!v(I).

Proof. Let r = c(I)+1 and suppose that g1 , . . . , gr is a nondegenerate

system of generators of I such that supp (g1 ) ∪ · · · ∪ supp (gr ) ⊆ Λ+ (I).
Let d : AI → N ∪ {+∞} be the filtration in AI induced by Γ+ (I).
Then, following the same argument as in the proof of Theorem 2.8, we
can show the acyclicity of the Koszul complex associated to the initial
forms of g1 , . . . , gr with respect to d. Then we apply [7, Lemma 2.9]
and [7, Theorem 4.1] to state that
AI
e({g1 , . . . , gr }AI , AI ) = dim C = r!v(I),
{g1 , . . . , gr }AI
508 C. BIVIÀ-AUSINA

where the first equality is a consequence of the fact that AI is Cohen-

Macaulay [6].
Suppose that g1 , . . . , gs is a nondegenerate system of generators of I
where s ≥ r. Then, following the idea of the proof of Theorem 2.10, we
can take a reduction J of {g1 , . . . , gs }AI in AI generated by a nonde-
generate system of r elements and constructed by linear combinations
of g1 , . . . , gs . From the monomorphism AI → A, we observe that JA
must also be a reduction of I in A. Thus, by Remark 2.5 we have that
Λ+ (I) = Λ+ (J) and v(I) = v(J).
By the fact that J is a reduction of {g1 , . . . , gs }AI in AI , we have
that e(J, AI ) = e({g1 , . . . , gs }AI , AI ). But
e(J, AI ) = r!v(J) = r!v(I),
by the previous step.
Now we show the converse. Since the ideal I1 ⊆ AI is nondegenerate
and of finite relative codimension, we can use the previous implication
to state that e(I1 , AI ) = r!v(I1 ) = r!v(I). If we suppose that
e({g1 , . . . , gs }AI , AI ) = r!v(I), then {g1 , . . . , gs }AI is a reduction of
I1 in A1 by Theorem 2.6. Therefore, I is a reduction of I1 and I is
nondegenerate by Proposition 4.2.

Remark 4.4. By an analogous method used in the proof of Corol-

lary 2.11, we can apply the previous theorem to say that an ideal
I ⊆ A of finite relative codimension is nondegenerate if and only if
I¯ is generated by those monomials xk with k ∈ Γ+ (I).

5. An effective method to compute the multiplicity of a

monomial ideal. As we shall see in this section, it is possible to
canonically construct a minimal reduction JI of any given monomial
ideal I ⊆ A of finite colength. Then we give a solution for the problem
of effectively computing the multiplicity of any monomial ideal I ⊆ A of
finite colength. We also refer the reader to [3] for a completely different
approach to the problem of computing the multiplicity of a monomial
ideal.

Theorem 5.1. Let I = xk1 , . . . , xks  ⊆ A be a monomial ideal of

finite colength and consider the polynomial fI = xk1 + · · · + xks . Let
 NONDEGENERATE IDEALS 509

JI denote the ideal of A generated by x1 (∂fI /∂x1 ), . . . , xn (∂fI /∂xn ).

Then
A
e(I) = dim C .
JI

Proof. By virtue of Theorem 2.6 we have to prove that J¯I = I. ¯ We

observe that Γ+ (JI ) = Γ+ (I). Then the problem reduces to prove
that the ideal JI is Newton nondegenerate, see Theorem 2.10 and
Corollary 2.11. Since the integral closure of I is equal to the integral
closure of the ideal generated by the vertexes of Γ+ (I), we can assume
that {k1 , . . . , ks } is equal to the set of vertexes of Γ+ (I).
Let ∆ be a face of Γ+ (I), suppose that ∆ is equal to the convex
hull of {k1 , . . . , kr } for some r ≤ s. We can attach to ∆ the structure
of a simplicial complex S such that the set of vertexes of S coincides
with {k1 , . . . , kr }, see Caratheodory’s theorem in [4, p. 139] for details.
Therefore, we can suppose that k1 , . . . , kr are linearly independent, in
particular r = dim ∆+1. In order to see that the ideal of A∆ generated
by (x1 (∂fI /∂x1 ))∆ , . . . , (xn (∂fI /∂xn ))∆ has finite codimension in A∆
it is enough to check, by [7, Theorem 6.2], that
     
∂fI ∂fI
x ∈ Cn : x 1 (x) = · · · = xn (x) = 0
∂x1 ∆
∂xn ∆

⊆ {x ∈ Cn : x1 · · · xn = 0},
for each ∆
∈ S. We observe that
   
∂fI ∂(fI )∆
xi = xi ,
∂xi ∆ ∂xi
for any i = 1, . . . , n. Therefore, if we write ki = (ki1 , . . . , kin ) for
i = 1, . . . , r, the above system of equations is translated into
k11 xk1 + · · · + kr1 xkr = 0
..
.
k1 x + · · · + kr x = 0.
n k1 n kr

Since k1 , . . . , kr are linearly independent, then xki = 0 for all

i ∈ {1, . . . , r}. But this means that some coordinate of x must be
zero.
510 C. BIVIÀ-AUSINA

Example 5.2. Let I ⊆ C[[x1 , . . . , x9 ]] be the ideal generated

by the monomials x21 , x22 , x23 , x34 , x35 , x36 , x47 , x48 , x49 , x5 x6 , x7 x8 x9 . Then
using Theorem 5.1 and, for instance, the program Singular [5] for the
computation of dim C A/JI , we find that e(I) = 6912.

REFERENCES

1. C. Bivià-Ausina, T. Fukui and M.J. Saia, Newton filtrations, graded algebras

and colength of nondegenerate ideals, Math. Proc. Cambridge Philos. Soc. 133
(2002), 55 75.
2. W. Bruns and J. Herzog, Cohen-Macaulay rings, Cambridge Stud. Adv. Math.,
vol. 39, Cambridge Univ. Press, Cambridge, 1993.
3. D. Delfino, A. Taylor, W.V. Vasconcelos, N. Weininger and R.H. Villarreal,
Monomial ideals and the computation of multiplicities, in Commutative Ring The-
ory and Applications, Lecture Notes in Pure and Appl. Math., vol. 231, Marcel
Dekker, New York, 2002, pp. 87 107.
4. D. Eisenbud, Commutative algebra with a view toward algebraic geometry,
Springer Verlag, Berlin, 1995.
5. G.M. Greuel, G. Pfister and H. Schönemann, Singular version 1.2 User manual
in Reports on computer algebra, Centre for Computer Algebra, University of Kaiser-
slautern, 1998. Available at: http://www.mathematik.uni-kl.de/˜zca/Singular
6. M. Hochster, Rings of invariants of tori, Cohen-Macaulay rings generated by
monomials, and polytopes, Ann. of Math. 96 (1972), 318 337.
7. A.G. Kouchnirenko, Polyèdres de Newton et nombres de Milnor, Invent. Math.
32 (1976), 1 31.
8. M. Lejeune and B. Teissier, Clôture intégrale des idéaux et equisingularité,
Centre de Mathématiques, Université Scientifique et Medicale de Grenoble, 1974.
9. H. Matsumura, Commutative ring theory, Cambridge Univ. Press, Cambridge,
1992.
10. D.G. Northcott and D. Rees, Reductions of ideals in local rings, Math. Proc.
Cambridge Philos. Soc. 50 (1954), 145 158.
11. D. Rees, a-transforms of local rings and a theorem on multiplicities of ideals,
Math. Proc. Cambridge Philos. Soc. 57 (1961), 8 17.
12. , Lectures on the asymptotic theory of ideals, London Math. Soc.
Lecture Note Ser., vol. 113, 1988.
13. M.J. Saia, The integral closure of ideals and the Newton filtration, J.
Algebraic Geom. 5 (1996), 1 11.
14. J.P. Serre, Algèbre locale. Multiplicités, Springer-Verlag, New York, 1989.
15. B. Teissier, Monômes, volumes et multiplicités, Introduction à la théorie des
singularités, II, Travaux en Cours 37, Hermann, Paris, 1988.
16. , Variétés polaires II. Multiplicités polaires, sections planes et condi-
tions de Whitney, Algebraic Geometry (Proc. La Rábida, 1981), Springer-Verlag,
New York, 1981.
 NONDEGENERATE IDEALS 511

17. E. Yoshinaga, Topologically principal part of analytic functions, Trans. Amer.

Math. Soc. 314 (1989), 803 814.

Departament de Matemàtica Aplicada, Universitat Politècnica de

València, Escola Politècnica Superior d’Alcoi, Plaça Ferrándiz i
Carbonell, 2, 03801 Alcoi (Alacant), Spain
E-mail address: carbivia@mat.upv.es